R&D Opportunities in Arc/Spark Optical Emission Spectrometry

July 1, 2011


Volume 26, Issue 7

Here are nine suggestions for improving this valuable technique.

A recent tutorial (1) prompted this commentary. Although generally considered a "mature" technology, arc/spark optical emission spectrometry continues to provide opportunities for spectrochemists. Here, we have listed a few suggestions for further improvement.

Arc/spark optical emission spectrometry (OES) is essential for the production of most metals and many of the products that use metals in their production. In the making of metals, a liquid sample is extracted from a furnace, quickly cooled, surfaced on a belt or disk sander, lathe, or milling machine, and analyzed in seconds for all elements of interest by arc/spark OES. That analysis informs the melter about what further refining is needed or what alloys to add to make the desired alloys for aluminum foil, aluminum for engine blocks, or low carbon steel for car hoods.

Before World War II, detectors did not exist to effectively support arc/spark optical emission spectrometers. Spectrochemical analyses were performed with spectrographs that photographed the spectrum of standards and samples, and the spectra were then measured using densitometers or microphotometers. Skilled spectroscopists could perform a multielement analysis in about 1 h, which was considerably faster than classical "wet" chemical methods.

World War II fostered the development of the photo-multiplier tube (PMT) as a radar detector. Applied Research Laboratories (Glendale, California) along with Alcoa (Pittsburgh, Pennsylvania) and Baird Associates (Cambridge, Massachusetts) working with Dow Chemical (Midland, Michigan) saw the potential of the PMT as a spectrometer detector and somewhat concurrently began developing direct reading arc/spark spectrometers. The first commercially available arc/spark OES system appeared circa 1947 and was a major technological advance for the metal-producing industries because the analysis time was reduced from approximately 1 h to less than 10 min. These new instruments came to be colloquially referred to as direct readers.

In the ensuing years, other improvements were made in arc/spark OES, including the development of vacuum optical systems allowing the determination of P, S, and better C; computers and microprocessors allowing faster and more precise analyses; improved excitation sources allowing better precision of analysis; lower detection limits; and high energy pre-spark (HEPS).

By 1990, a relatively unskilled operator could perform a complete multielement analysis of any major metal in about 30 s, including sample preparation. In fact, the whole analytical process could be automated so that an operator was not needed for routine spectrometer operation.

At that time arc/spark OES was considered a mature technology and generally unworthy of the attention of the spectrochemical research community. There have always been too many other exciting avenues to pursue: inductively coupled plasma (ICP) spectroscopy, ICP–mass spectrometry (MS), nuclear magnetic resonance (NMR) spectroscopy, Fourier-transform infrared (FT-IR) spectroscopy, gas chromatography (GC)–MS, and many others. (Note: ICP users might be interested to know that the first ICP-OES system used for analyses was an Applied Research Laboratories [ARL, now part of Thermo Fisher Scientific, Austin, Texas] arc/spark OES system retrofitted with an ICP excitation source by Stanley Greenfield at the Albright and Wilson R&D Centre in the United Kingdom circa 1961.)

However, this viewpoint should have changed significantly with the advent of several new technologies developed between 1985 and 2000:

  • proved spark creation

  • Charge-coupled device (CCD) optics

  • Gas-filled vacuum ultraviolet (VUV) optical chamber

  • Digitally controlled arc/spark excitation sources

  • Advanced electronic time-resolved spectroscopy (TRS)

  • Single-spark evaluation and the mapping of inclusions in metals

Any one of these new technologies should have produced a flurry of papers. Instead, the research was mainly carried out in industrial laboratories by manufacturers of OES instrumentation, primarily Spectro Analytical Instruments (now part of Ametek, Berwyn, Pennsylvania) and ARL. Few papers were published.

We suggest that significant opportunities still exist to improve analyses by arc/spark OES. Some have been listed below.

Improved Analytical Stability by Revisiting the Excitation Source

The spark excitation research conducted by chief engineer Kasper Habosian at Baird circa 1975–1985 was based on previous work done by John Walters and his graduate students at the University of Wisconsin (Madison, Wisconsin). Habosian replaced the resistor in the resistor–capacitor (RC) circuits of the excitation source with an inductor (coil) that did not heat or change characteristic value with use. Over long-term use a resistor heats and changes its resistance, however slightly. This change in resistance changes the excitation source conditions, which negatively affect analytical stability. A coil in place of the resistor serves the same role in spark creation, but it does not heat with use and thereby improves long-term stability.

Habosian's further work included a smaller coil and capacitor, which created a spark with an extremely short rise rime (≤2 µs), which obviated much, but not all of the spectral background (2). The sparks created with this circuitry were extremely stable spatially and spectral lines did not waver in their exit slits. Therefore very fast rise time sparks are ideal for TRS studies. It would be interesting to revisit this apparently "lost" technology.

The relatively new digitally controlled excitation sources present numerous avenues for research. A systematic exploration of the new excitation options available is indicated.

Improved Analysis with Homologous Spectral Lines

Investigations of homologous spectral lines, both experimental and theoretical, beg to be undertaken. The basic definition of homologous spectral line pairs, due to Gerlach (3), is that this ratio of line intensities serves to compensate for changes in excitation energy. A recent introduction to this subject is available (4). Some suggested experiments, along with results from the early 1990s, may also be found (5).

In most good PMT-based arc/spark OES systems used in the analysis of steel the line pair of Cr(II) 267.7 nm ratio to Fe(II) 271.4 nm routinely produce excellent relative standard deviations (RSD) of 0.2–0.3%. Other elements of importance produce worse RSDs because they are not as homologous with the Fe 271.4-nm internal standard.

Both atomic and ionic internal standards have been used only since the introduction of optical fibers as light paths to multiple optics, an innovation due to Spectro Analytical Instruments (this approach dates back to the early-to-mid 1980s). However, now with CCD and CID (charge induced device) based arc/spark OES the complete spectrum is available without the added cost or space needed for multiple internal standard channels. Therefore, a study of superior internal standards for all elements of interest is available without additional hardware. The study would require considerable work and some software support for the voluminous line comparisons required. It might be beyond the resources of arc/spark OES system producers. However, industry and academia collaborations appear to be the order of the day in the pharmaceutical industry. Why not in spectrochemistry?

Theoretically, a set of necessary and sufficient conditions should be developed. Similar excitation potentials are not enough (5).

Improved Stability Using Fixation Pairs

There is also the issue of fixation pairs of spectral lines, another concept introduced by Gerlach and the opposite of homologous spectral line pairs. Generally, we choose an atomic and ionic spectral line of the matrix or base element, although spectral lines of any significant analyte may be used. The ratio of the two will provide a sensitive indicator of any changes in excitation source conditions.

In ICP-OES the Mg(II) (280.2 nm)/Mg(I) (285.2 nm) intensity ratio has been used to monitor the "robustness" of the plasma (6). This means a capability of dealing with the introduction of a complex matrix aerosol without significantly changing the plasma physical properties (for example, the excitation conditions).

This concept is likely applicable to modern arc/spark spectrometers for improved short- and long-term precision.

Improved Analysis with New Spectral Lines

Many new spectral lines became useable for both arc/spark and ICP-OES systems in the mid-1990s because of the development of the nitrogen gas-filled VUV optical chamber. In arc/spark the new applications included gases in metals (H 121.6 nm, N 149.2 nm, O 130.2 nm) and ultralow carbon (165.8 nm and 133.4 nm). In ICP-OES, the early applications involved the halogens (Cl 134.7 nm, Br 157.4 nm, I 161.7 nm) and Al 167.1 nm (7,8).

It should be noted that vacuum optical systems cannot be used for lines below a wavelength of about 180 nm because of polymerization of the entrance optic, a window or lens. In the best vacuum optical systems some 1013 molecules still exist and many are hydrocarbons from the vacuum pump oil, outgassing of electronic insulation, and so forth. They exist even with subzero and other traps above the pump. These hydrocarbons drift about in the optical chamber, and when they impinge on the entrance optic during sample excitation, the extreme light causes them to adhere to the lens or window as some sort of light brown film or polymer. The coating has little effect on lines above 190 nm, but has progressive absorption below 180 nm, to the point that lines below are completely unusable. ARL has had some success flushing the inside of the entrance optic on their vacuum spectrometer with argon for the analysis of nitrogen at 149.2 nm, but there still appear to be questions regarding long-term stability.

Considerable work has been done in the past decade (9). Will there be other spectral lines of interest discovered? Almost certainly.

Improved Analysis with Time-Resolved Spectroscopy

In spark creation much of the spectral background occurs early in the spark with the excitation of the elements of interest occurring at various times later during the spark. Moreover, the excitation of interfering elements may occur at different times during the spark. Current electronics provide a way to gate signal response (1).

Time-resolved spectroscopy provides several benefits for analysts: improved detection limits for atomic spectral lines; reduced interelement effects from ionic spectral lines; and improved calibration curves.

What are the optimal time delays (from spark initiation) and integration times for each spectral line of the elements? Are there other advantages not yet discovered?

Improved Understanding of Matrix Effects

A first pass was made by Slickers in his book (10), listing several factors contributing to the "matrix effect." However, it is unlikely that this is all there is to say on the subject. An exploration of root causes and links between the various factors listed by Slickers is a suitable research topic. A recent tutorial on the subject of interelement corrections is available (11).

A Basic Theoretical Issue

The way we make line overlap corrections is fundamentally flawed. In the arc/spark calibration and measurement process, the first step is to measure the absolute intensities of the analyte (IEl) and internal standard (IS) spectral lines. Next, the intensity ratios are computed (IEl/IS). Calibration curves are generated in the form of intensity ratios versus concentration ratios. Now, we finally make line overlap corrections, but subtract intensity ratios. Theoretically, we should subtract the overlap line intensity from the analyte intensity before the ratio process. It makes no sense to subtract intensity ratios and yet the existing process works quite well. Why?

Fundamental Parameters Approach to Calibration in OES

In X-ray fluorescence (XRF) spectrometry it is possible to generate calibration curves based solely on the "fundamental parameters" of the physics involved. The possibility of doing the same with OES was addressed in a recent tutorial (12): "Will we ever be able to compute theoretical calibrations in optical emission spectrometry? Certainly a theoretical framework exists for optical emission spectrometry." The author went on to explain that progress has been difficult due to the richness of the spectra coupled with the complexity of the excitation sources. Nevertheless some important work has already been done (13–15).

Improved Handheld Arc/Spark OES Systems

The two existing commercially available handheld arc/spark instruments that we are aware of both use arc-in-air excitation and show limited analytical capability. They are basically metal "sorters." This is in sharp contrast to several handheld XRF units, which can produce analytical results rivaling those of benchtop and laboratory energy dispersive XRF instruments.

The use of a small nitrogen-filled optical system with the ability to use the carbon lines at 165.8 nm and 133.4 nm, and the addition of a small, lightweight argon cylinder would allow spark-like discharges in an argon atmosphere, eliminating some of the disadvantages of the arc-in-air discharge, and perhaps allow the identification of the "L grade" (low carbon) stainless steels.


Despite its status as a "mature" technology, arc/spark spectrometry — a workhorse of the metals industry — continues to present innovative technology enabling new and exciting applications. Some ideas for future R&D in this area were presented here.


(1) V. Thomsen, Spectroscopy 25(10), 42–45 (2010).

(2) J.J. Fox and J.L. Spencer, "Analytical Performance of Present Day Direct-Reading Spectrometers," Baird Corp. Applications Report #790423-70, circa 1980.

(3) W. Gerlach and E. Schweitzer in Foundations and Methods of Chemical Analysis by the Emission Spectrum (Hilger, London, England, 1929). English translation of Die chemische Emissionsspektralanalyse, Volume One (Voss, Leipzig, 1929).

(4) V. Thomsen, Spectroscopy 17(12), 117–120 (2002).

(5) V. Thomsen, "Modern Spectrochemical Analysis of Metals: An Introduction for Users of Arc/Spark Instrumentation," presented at ASM International, Materials Park, Ohio, 1996.

(6) J.M. Mermet and E. Poussel, App. Spec., Focal Point 49(10), 12A–18A (1995).

(7) V. Thomsen, G. Roberts, and D. Tsourides, Amer. Lab. 29(16), 18H–M (1997).

(8) UV-PLUS Spectrometer, Spectro Analytical Instruments, U.S. Patent 5225681.

(9) P. Heitland, "Vacuum Ultraviolet (VUV) Wavelength Coverage for ICP-OES," Spectro Analytical Instruments Applications Report, circa 2005.

(10) K. Slickers, Automatic Emission Spectroscopy (Bruehlische Univ. Press, Second Edition, 1993).

(11) V. Thomsen, D. Schatzlein, and D. Mercuro, Spectroscopy 21(6), 32–40 (2006).

(12) V. Thomsen, Spectroscopy 22(5), 46–50 (2007).

(13) A. Bogaerts, R. Gijbels, and J. Vleck, Spectrochimica Acta B 53, 1517–1526 (1998).

(14) A. Bogaerts and R. Gijbels, J. Anal. Atomic Spectrom. 13, 721–726 (1998).

(15) A. Bogaerts, Z. Chen, R. Gijbels, and A. Vertes, Spectrochimica Acta B 58, 1867–1893 (2003).

Volker B.E. Thomsen a physicist by training, has some 30 years of experience in elemental spectrochemical analysis (OES and XRF). He is currently a consultant in this area from his home in Atibaia, São Paulo, Brazil. His other interests include mineralogy and history of science. Occasionally, he still plays the blues harmonica. He can be reached at vbet1951@uol.com.br.

Volker B.E. Thomsen

Jerald L. Spencer is now retired. However, while employed he was the Director of Marketing for Baird Corporation when it was a major arc/spark OES supplier; Spectro Analytical Instruments GmbH when it was the largest arc/spark OES supplier; and Leeman Labs, Inc. For 10 years before retirement, he was an independent marketing consultant serving a number of analytical instrument companies, including two major arc/spark emission spectrometer companies.